Note: Sorry for delays today. My host was in maintenance mode.
Fed’s Beige Book Shows Economic Activity Up Modestly
US economic activity increased modestly in the past two months with demand from data centers, in particular, driving growth, the Federal Reserve said.
The outlook for the economy was “positive,” according to the US central bank’s Beige Book survey of regional business contacts released Wednesday, though sentiment was mixed across sectors amid uncertainty about energy prices and geopolitics.
While spending on high-end purchases was solid, the report also noted increased price sensitivity. Manufacturing activity grew across most of the Fed’s districts on the back of demand for defense and data-center orders. Employment rose slightly across the country. (…)
Prices, meanwhile, accelerated moderately in most districts.
“Consumer-facing contacts in a few districts noted that heightened price sensitivity among customers was putting a limit on their ability to pass through input price increases,” the report said. (…)
But the Atlanta Fed’s GDP Now has Q3 GDP up 4.8%!
New York Fed President John Williams yesterday echoed Scott Bessent: rising Treasury yields are driven by “a strong US economy and a strong economic outlook fueled by big investments in AI and data centers and technology in general.”
AI-related growth is offsetting whatever weaknesses there are. Note that this chart has a single scale.
One of the problem with AI growth is that its urgency makes it totally insensitive to prices and financing costs. Nvidia is also acting as the central banker of the AI economy.
Williams rightly said there are “no clear signs right now” whether current policy is sufficient to return inflation to target.
There are also no clear signs that monetary policy actually matters nowadays. Read on.
The Bond Market’s Signal Is About to Get Louder More aggressive rate hikes will be needed to tame inflation and bond yields that have yet to peak.
The Treasuries market is flashing a warning signal: the US economy has become increasingly insensitive to interest rate increases by the Federal Reserve. To combat inflation and tame yields on long-dated debt, more aggressive hikes will be needed. That means bonds will keep losing value as yields have yet to peak.
- The transmission of monetary policy has become slower and more uneven, with pockets of acute vulnerability in consumer credit and corporate debt too.
- In this environment, modest interest rate increases fall short, allowing inflation to be sticky enough to push measures of long-term price expectations higher.
- Yields on long-dated Treasuries, that peaked around 5% in the last hiking cycle, could push even higher this time.
- While equities are resilient as earnings and spending grow, the risk of more Fed rate hikes acts as an overhang on the market.
(…) The single largest structural change is the dominance of long-term fixed-rate mortgages. In the 1980s, adjustable-rate mortgages were far more prevalent. That meant Fed hikes transmitted almost immediately to household budgets. Today, the vast majority of US homeowners hold 30-year fixed-rate mortgages. And since many of those were refinanced at historically low rates during 2020 and 2021, debt-servicing costs remained around 10% of income despite the 2022–2023 hiking cycle.
On the corporate side, it’s similar. In the 1980s, corporate America carried more floating-rate bank debt and had less access to deep, long-duration bond markets. Investment-grade and high-yield bond markets since then have allowed companies to lock in long-term fixed-rate financing, reducing their immediate exposure to rate moves. (…)
The pain of higher rates was meted out, then, to lower-income borrowers via credit cards and auto loans. That produced a K-shaped outcome, or a so-called ‘vibecession,’ which in parts of the economy was very real. The double whammy of inflation and higher interest rates disproportionately impact lower-income households and small businesses. (…)
Thanks to AI spending, recession is even less of a concern this time around. (…)
- One important point is that this is a global selloff, which makes it hard for the US to buck the trend.
- No one knows where the tail risks are yet. They could be in Japan, where intervention is ongoing.
- Meanwhile, the bond market’s gains after US Treasury buyback plans were announced have evaporated.
- The potential for the Iran War to last into 2027 heightens the risks.
Add urgent military spending, urgent green spending, urgent supply chain spending, all price insensitive.
But more and more Americans are price sensitive:
- The owner of the Circle K brand reported fuel revenues of $16.7 billion in its fiscal first quarter, up 33% from the same period last year. Same-store fuel volumes fell by 1.6% in the US and 4.3% in Europe and other regions, and increased by 1.1% in Canada. Same-store merchandise revenues rose by 1.7% or less across all markets in the period ended July 19, largely missing estimates from analysts surveyed by Bloomberg, and well below inflation.
- The 60+ day delinquency rate on US subprime auto loans is up to ~5.2%, the highest on record. This figure has more than doubled over the last 4 years. Serious delinquency rates on subprime auto loans are now ~1.7 percentage points above their 2008 Financial Crisis peak. At the same time, 60+ day delinquencies on prime auto loans are up to ~0.4%, near their highest since 2011. Meanwhile, total US auto debt surged +$28 billion in Q2 2026, to a record $1.71 trillion. Americans are falling behind on their car payments at a historic rate. (@KobeissiLetter)
One offset:
America’s Population of 401(k) Millionaires Keeps Growing, Buoyed by Markets
The number of millionaire 401(k) accounts at Fidelity Investments rose 19% to a record 769,000 between the first and second quarter, according to a report released Thursday. It was the largest quarterly increase since the fourth quarter of 2023, the company said.
Aiding savers was a blockbuster quarter for equities. The S&P 500 Index gained about 15% in the three months ended June 30, its strongest performance since 2020. The average 401(k), 403(b) and IRA account balances on Fidelity’s platform rose to all-time highs, while savings rates for workplace retirement plans also held at record levels, the company said. (…)
Retirement savers are hitting the landmark even while many report feeling underprepared for their later years. The share of workers who say they feel confident about having enough money to live comfortably throughout retirement fell to the lowest level since 2017, according to a joint Retirement Confidence Survey from the Employee Benefit Research Institute and Greenwald Research released earlier this year.
Debt, inflation and rising housing and healthcare costs are hampering savings plans, according to the research. Others are worried about the future of Social Security. New projections from June estimate that the Social Security Trust Fund may be depleted by 2032.
Estimates vary widely on how much people need to save for retirement. The size of that nest egg depends on where they live, their expenses, financial goals and desired standard of living. Americans say they need $1.46 million on average to retire comfortably, according to Northwestern Mutual’s 2026 Planning & Progress Study. (…)
“It’s maybe not as big a deal to be a millionaire as it might’ve been when you watch Gilligan’s Island in the ‘60s,” he said. “The millionaire was a rich person. Now, it just doesn’t go as far as it used to.” (…)
At some point, interest rates will start to bite.
- On spending
- On margin debt
Margin debt, as it has during other speculative periods, is growing considerably faster than either credit card debt or mortgage debt. Maybe the Federal Reserve (Fed) should consider hiking margin requirements instead of the fed funds rate? (RBA)
- On asset allocation. 10Yr yields at 5%+ with inflation below 3% and a resolutely (?) hawkish Fed could become more widely appealing.
Especially if productivity offsets other inflationary pressures:
Dell Results Suggest AI Productivity Boom Is Here
Dell Technologies’ stock price is soaring. The company delivered a major beat across the board for its fiscal 2027 second quarter (ended July 31), driven by massive, accelerating demand for AI infrastructure and strong legacy hardware performance.
Revenues and earnings rose 58% y/y and 203%, respectively. AI server revenue rose 100%, while traditional servers and networking revenues rose 122%.
The results confirm that the AI infrastructure buildout remains in full swing. Strong demand for AI compute capacity points to accelerating AI adoption across the economy, which we think will drive a productivity boom. (…)
Productivity growth has rebounded since it last bottomed in Q2-2017 at 0.85%, based on the annualized average of its seven-year growth rates. It rose to 2.4% during Q2-2026, slightly exceeding its historical average of 2.3%. We predict that this growth rate will rise to 3.0%-4.0% by the end of the decade.
From the NY Fed:
AI Adoption Has Become Much More Widespread in the Workplace
Our August business surveys asked firms in the New York and Northern New Jersey region whether they used AI as part of their business processes in the past six months, questions we have asked each year since 2024.
AI adoption in the workplace has continued to increase sharply and has now become widespread. As shown in the chart below, 61 percent of service firms reported using AI this year, up from 40 percent last year and 25 percent in 2024.
Businesses in knowledge-intensive sectors, such as information, business services, and finance, had the highest usage rates. Among manufacturers, 51 percent reported using AI as part of their business processes, roughly double the 26 percent from last year and triple the 16 percent in 2024. These shares are toward the high end of the range of existing studies of AI use in the workplace.
While AI adoption has become widespread, most firms have made only limited investments in the technology. Three-quarters of service firms and more than 90 percent of manufacturers characterize their AI investments as minimal to modest, ranging from use of free AI tools to allocating a small share of overall spending to AI tools or services.
Meanwhile, just 15 percent of service firms—but no manufacturers—indicate they have committed significant resources to AI adoption, with only about 5 percent of service firms characterizing AI adoption as a major strategic investment. Among AI adopters, the median share of workers using it was just 17 percent for service firms and 7 percent for manufacturers.
In short, AI adoption in the workplace is now fairly broad but investments and worker usage remain limited.
With AI use in the workplace now widespread, why have some businesses refrained from adopting it?
(…) cost does not seem to be the main deterrent—it was among the least cited reasons by non-adopters. About half of non-adopters said the type of work they do does not lend itself to AI, while roughly a quarter indicated AI is currently not good enough to provide benefits to their business.
There were also some concerns about using AI. More than a third of non-adopters were concerned about data privacy, security, or confidentiality, and a similar share expressed concerns about accuracy or reliability. Further, roughly a third indicated they currently lack staff with the technical skills to use it effectively. (…)
Consistent with our earlier surveys, existing workers are much more likely to be retrained than replaced by AI. Among businesses that use AI, just over a third of service firms and more than 20 percent of manufacturing firms report retraining workers in response to AI. Firms report retraining workers across the educational spectrum, though somewhat more of those with college degrees.
These findings align with the broader research literature, which also tends to find limited labor market effects from AI adoption so far in terms of layoffs or reduced hiring. However, one recent study suggests entry-level workers may be affected significantly, as AI can substitute for routine tasks often performed by newer employees, potentially creating barriers to workforce entry even as it enhances productivity for experienced workers. (…)
Evidence from our surveys so far confirms what many studies are showing: that AI has been more likely to augment workers than replace them.
The most striking number in Dell’s release was that Dell’s AI-optimized server revenue came in at $16.4 billion, up 100% year over year. Crucially, the AI server business had a record $95 billion backlog as of the end of the second quarter. Companies are rapidly equipping for AI.
Global data center spending is set to reach $31.6 trillion through 2050 to meet the world’s growing appetite for AI, an investment boom with no precedent in history, according to PricewaterhouseCoopers LLP.
Dwarfing projects such as the railways, internet and electrification, spending on data centers could even hit $50 trillion over the next two and a half decades if AI adoption accelerates beyond PwC’s “central scenario” forecast, the firm said in a report Wednesday. For comparison: the US gross domestic product is roughly $30 trillion. (…)
The bulk of the spending will go into what fills the data centers — hardware from companies such as global AI chip leader Nvidia Corp. (…)
Spending will keep rising through mid-century as graphics processing units, servers, storage systems, networking equipment and other hardware will require routine replacement. Recurring chip upgrades — the computational power — and not land or construction, will account for most of the investment, quite unlike traditional capex cycles like prior generations of memory chip production or the global fiber internet rollout, which “front loaded” investments, taking on costs and risks upfront. (…)
On an annual basis, global data center spending will increase from about $800 billion this year to $1.1 trillion in 2030 and $1.8 trillion in 2050, PwC predicted. China and India will drive the largest share of incremental demand, supported by large populations, rapidly expanding digital economies, and substantial headroom for AI to embed in business and consumer activity. (…)
While global demand is strong, factors such as power availability, data sovereignty requirements and the flow of semiconductors will determine which regions capture the investments, PwC said. Power will be the foremost factor that shapes where AI infrastructure investment occurs.
Indeed, much of the forecast hinges on how fast reliable electricity supply for data centers can be established, according to the report. Affordable, reliable, and increasingly low-carbon electricity at scale is the hardest requirement for many markets to meet.
And while the researchers’ projection assumes a fairly open trading system where chips move freely across borders, disruptions in semiconductor supply chains could cut global investment by nearly 20%, they said. Meanwhile, a growing sovereignty push could redistribute, but not reduce, global investment.
“The $31.6 trillion question isn’t whether the capital exists. It does,” the researchers said. “Nor is the question whether the demand is real. It is. The question is which regions, operators, and institutions are positioned to capture it and which aren’t.”
Elon Musk Monday warned that the artificial intelligence industry is racing toward an imminent global power crisis, predicting a massive 15-gigawatt energy shortfall by 2027. Musk said that electricity has officially replaced chips as the primary bottleneck for AI development.
He revealed that AI deployment is growing exponentially at 40% to 50% annually, while power capacity outside of China is crawling forward at just 10% to 20% per year.
Without a rapid intervention in power infrastructure, Musk warned that billions of dollars in advanced AI processors will sit completely idle.
Musk noted that while China possesses substantial electricity infrastructure, strict GPU export bans limit their chip access. Conversely, Western tech hubs have the chips but lack the raw wattage to support them
I bet it will be easier for China to solve its chip problem than for the US its power challenges.